Dear Guillaume,
The question is not whether 2.14 is outdated, but whether the bug has
miraculously fixed itself in later versions.
> $ time perl -e '$|=0; foreach $x (1..500) { print "x" foreach(1..$x);
> print " $x\n" }'
> (snip)
> real 0m0.458s
> on Intel(R) Core(TM)2 CPU T7400 @ 2.16GHz
I am not impressed. The bug better exhibits itself with $|=1.
Testing my ancient "Intel(R) Pentium(R) 4 CPU 2.40GHz stepping 04" with
"MSI 6526G-L with 845G chipset" motherboard and using the builtin video
controller, running Ubuntu hardy with i810 driver, I get:
Running rxvt:
p...@asi:~$ ps --no-headers -p $PPID
5701 pts/2 00:00:03 rxvt
p...@asi:~$ time perl -e '$|=0; foreach $x (1..500) { print "x" foreach(1..$x);
print " $x\n" }'
real 0m0.064s
and that may be a good benchmark (anything slower being unacceptable?).
Even rxvt with flushed output is bad:
p...@asi:~$ ps --no-headers -p $PPID
5701 pts/2 00:00:03 rxvt
p...@asi:~$ time perl -e '$|=1; foreach $x (1..500) { print "x" foreach(1..$x);
print " $x\n" }'
real 0m0.509s
Trying xterm, is not impressive:
p...@asi:~$ ps --no-headers -p $PPID
5624 ? 00:00:08 xterm
p...@asi:~$ time perl -e '$|=0; foreach $x (1..500) { print "x" foreach(1..$x);
print " $x\n" }'
real 0m0.683s
p...@asi:~$ time perl -e '$|=1; foreach $x (1..500) { print "x" foreach(1..$x);
print " $x\n" }'
real 0m0.766s
while gnome-terminal is hopeless:
p...@asi:~$ dpkg -l gnome-terminal
ii gnome-terminal 2.22.1-0ubuntu The GNOME 2 terminal emulator application
p...@asi:~$ ps --no-headers -p $PPID
5637 pts/1 00:00:04 gnome-terminal
p...@asi:~$ time perl -e '$|=0; foreach $x (1..500) { print "x" foreach(1..$x);
print " $x\n" }'
real 0m0.932s
p...@asi:~$ time perl -e '$|=1; foreach $x (1..500) { print "x" foreach(1..$x);
print " $x\n" }'
real 0m1.184s
Please comment on why does flushed output take so long in the X server,
and why is gnome-terminal so inefficient. Actually improving them would
be appreciated.
Please re-open bug.
Thanks, Paul
Paul Szabo p...@maths.usyd.edu.au http://www.maths.usyd.edu.au/u/psz/
School of Mathematics and Statistics University of Sydney Australia
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
